1 Análisis PPM-S

2 Wave 01

library(sjPlot)
library(dplyr)
library(lavaan)
library(stargazer)
library(corrplot)
library(psych)
library(knitr)
library(kableExtra)

data01 <- sjlabelled::read_spss(path = "input/data/original/Estudio_3_ola1.sav",verbose = FALSE)

dat01 <- data01 %>% filter(Intro==1,Finished==1) %>% select(starts_with("meritv01")) %>% na.omit()
dat02 <- data01 %>% filter(Intro==1,Finished==1) %>% select(starts_with("meritv02")) %>% na.omit()
dat03 <- data01 %>% filter(Intro==1,Finished==1) %>% select(starts_with("meritv03_p")) %>% na.omit()

# - Nos quedamos solamente con los casos que ACEPTARON participar y terminaron el cuestionario.
# - Para dejar una base que contenga solamente los casos que tienen información completa en los ítems usamos na.omit()

dat04 <- data01 %>% filter(Intro==1) %>%
  select(starts_with("meritv01"),
         starts_with("meritv02"),
         starts_with("meritv03_p"),
         starts_with("FL_21_DO")) 
dat04$grupo <- NA
dat04$grupo[dat04$FL_21_DO_merit_perc_pref_julio19v01==1] <- 1
dat04$grupo[dat04$FL_21_DO_merit_perc_pref_julio19v02==1] <- 2
dat04$grupo[dat04$FL_21_DO_merit_perc_pref_julio19v03==1] <- 3
dat04$perc_effort <- rowSums(dat04[,c(matches(match = "perc_effort",vars = names(dat04)))],na.rm = TRUE)
dat04$perc_talent <- rowSums(dat04[,c(matches(match = "perc_talent",vars = names(dat04)))],na.rm = TRUE)
dat04$perc_wpart  <- rowSums(dat04[,c(matches(match = "perc_wpart" ,vars = names(dat04)))],na.rm = TRUE)
dat04$perc_netw   <- rowSums(dat04[,c(matches(match = "perc_netw"  ,vars = names(dat04)))],na.rm = TRUE)

dat04$pref_effort <- rowSums(dat04[,c(matches(match = "pref_effort",vars = names(dat04)))],na.rm = TRUE)
dat04$pref_talent <- rowSums(dat04[,c(matches(match = "pref_talent",vars = names(dat04)))],na.rm = TRUE)
dat04$pref_wpart  <- rowSums(dat04[,c(matches(match = "pref_wpart" ,vars = names(dat04)))],na.rm = TRUE)
dat04$pref_netw   <- rowSums(dat04[,c(matches(match = "pref_netw"  ,vars = names(dat04)))],na.rm = TRUE)
dat04[dat04==0] <- NA

2.1 Version 01: Fixed order by perception / preferences (N=697)

Percepcion esfuerzo
Percepcion talento
Percepcion familia rica
Percepcion redes
Preferencia esfuerzo
Preferencia talento
Preferencia familia rica
Preferencia redes


Statistic	N	Mean	St. Dev.	Min	Pctl(25)	Pctl(75)	Max

meritv01_perc_effort	697	3.164	1.360	1	2	4	5
meritv01_perc_talent	697	2.981	1.122	1	2	4	5
meritv01_perc_wpart	697	3.603	1.353	1	3	5	5
meritv01_perc_netw	697	3.740	1.255	1	3	5	5
meritv01_pref_effort	697	3.908	1.225	1	3	5	5
meritv01_pref_talent	697	3.268	1.168	1	3	4	5
meritv01_pref_wpart	697	2.555	1.117	1	2	3	5
meritv01_pref_netw	697	2.337	1.057	1	1	3	5

cor01<-polychoric(dat01)
corrplot(cor01$rho,method = "color" ,type = "upper", order = "hclust", tl.col = "black", tl.srt = 45,addCoef.col = "black",diag = FALSE)

model01 <- '
perc_merit=~meritv01_perc_effort+meritv01_perc_talent 
perc_nmerit=~meritv01_perc_wpart+meritv01_perc_netw  
pref_merit=~meritv01_pref_effort+meritv01_pref_talent 
pref_nmerit=~meritv01_pref_wpart+meritv01_pref_netw'

fit1_c <- cfa(model = model01,data = dat01,estimator="MLR") # Continuo/ estimador ML Robust

fit1_o <- cfa(model = model01,data = dat01,
            ordered = c("meritv01_perc_effort","meritv01_perc_talent",
                        "meritv01_perc_wpart","meritv01_perc_netw",
                        "meritv01_pref_effort","meritv01_pref_talent",
                        "meritv01_pref_wpart","meritv01_pref_netw"))

#---------------------------------------------------#
# #summary(fit1_c,standardized=TRUE, fit.measures=TRUE)}
fitmeasures(fit1_c)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]

## chisq.scaled           df          cfi   cfi.scaled   cfi.robust 
##  42.93480226  14.00000000   0.97948028   0.97736086   0.98021257 
##        rmsea rmsea.scaled rmsea.robust 
##   0.05812794   0.05445401   0.05695600

semPlot::semPaths(object = fit1_c,what = "std",thresholds = FALSE, intercepts = FALSE,fade=FALSE, curve = 1.5)

# #summary(fit1_o,standardized=TRUE, fit.measures=TRUE)
fitmeasures(fit1_o)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]

## chisq.scaled           df          cfi   cfi.scaled   cfi.robust 
##  46.11414265  14.00000000   0.99763575   0.99207488           NA 
##        rmsea rmsea.scaled rmsea.robust 
##   0.03794574   0.05740894           NA

semPlot::semPaths(object = fit1_o,what = "std",thresholds = FALSE, intercepts = FALSE,fade=FALSE, curve = 1.5)

#summary: good fit, heywood case item 8.

2.2 Version 02: Fixed order, by topic (N=706)

Percepcion esfuerzo
Preferencia esfuerzo
Percepcion talento
Preferencia talento
Percepcion familia rica
Preferencia familia rica
Percepcion redes
Preferencia redes


Statistic	N	Mean	St. Dev.	Min	Pctl(25)	Pctl(75)	Max

meritv02_perc_effort	706	3.105	1.399	1	2	4	5
meritv02_pref_effort	706	3.741	1.339	1	3	5	5
meritv02_perc_talent	706	3.035	1.111	1	2	4	5
meritv02_pref_talent	706	3.146	1.149	1	2	4	5
meritv02_perc_wpart	706	3.661	1.364	1	3	5	5
meritv02_pref_wpart	706	2.742	1.209	1	2	4	5
meritv02_perc_netw	706	3.728	1.218	1	3	5	5
meritv02_pref_netw	706	2.385	1.109	1	1	3	5

cor02<-polychoric(dat02)
corrplot(cor02$rho,method = "color", type = "upper", order = "hclust", tl.col = "black", tl.srt = 45,addCoef.col = "black",diag = FALSE)

model02 <- '
perc_merit=~meritv02_perc_effort+meritv02_perc_talent 
perc_nmerit=~meritv02_perc_wpart+meritv02_perc_netw  
pref_merit=~meritv02_pref_effort+meritv02_pref_talent 
pref_nmerit=~meritv02_pref_wpart+meritv02_pref_netw'

fit2_c <- cfa(model = model02,data = dat02,estimator="MLR") # Continuo/ estimador ML Robust

fit2_o <- cfa(model = model02,data = dat02,ordered = c("meritv02_perc_effort","meritv02_perc_talent",
                                                     "meritv02_perc_wpart","meritv02_perc_netw",
                                                     "meritv02_pref_effort","meritv02_pref_talent",
                                                     "meritv02_pref_wpart","meritv02_pref_netw"))

#---------------------------------------------------#
#summary(fit2_c,standardized=TRUE, fit.measures=TRUE)
fitmeasures(fit2_c)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]
semPlot::semPaths(object = fit2_c,what = "std",thresholds = FALSE, intercepts = FALSE,fade=FALSE, curve = 1.5)

#summary(fit2_o,standardized=TRUE, fit.measures=TRUE)
fitmeasures(fit2_o)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]
semPlot::semPaths(object = fit2_o,what = "std",thresholds = FALSE, intercepts = FALSE,fade=FALSE, curve = 1.5)

## chisq.scaled           df          cfi   cfi.scaled   cfi.robust 
## 100.06185323  14.00000000   0.92567916   0.91208862   0.92577394 
##        rmsea rmsea.scaled rmsea.robust 
##   0.09562704   0.09331232   0.09530656 
## chisq.scaled           df          cfi   cfi.scaled   cfi.robust 
## 101.48249122  14.00000000   0.98421904   0.96229187           NA 
##        rmsea rmsea.scaled rmsea.robust 
##   0.07128592   0.09414603           NA

NOTA: bad fit, heywood en estimación DWLS

2.3 Version 03: Random order (N=697)


Statistic	N	Mean	St. Dev.	Min	Pctl(25)	Pctl(75)	Max

meritv03_perc_effort	697	3.329	1.370	1	2	5	5
meritv03_perc_talent	697	3.072	1.245	1	2	4	5
meritv03_perc_wpart	697	3.722	1.360	1	3	5	5
meritv03_perc_netw	697	3.900	1.235	1	3	5	5
meritv03_pref_effort	697	4.046	1.159	1	4	5	5
meritv03_pref_talent	697	3.339	1.229	1	2	4	5
meritv03_pref_wpart	697	2.770	1.189	1	2	4	5
meritv03_pref_netw	697	2.506	1.144	1	2	3	5

cor03<-polychoric(dat03)
corrplot(cor03$rho,method = "color", type = "upper", order = "hclust", tl.col = "black", tl.srt = 45,addCoef.col = "black",diag = FALSE)

model03 <- '
perc_merit=~meritv03_perc_effort+meritv03_perc_talent 
perc_nmerit=~meritv03_perc_wpart+meritv03_perc_netw  
pref_merit=~meritv03_pref_effort+meritv03_pref_talent 
pref_nmerit=~meritv03_pref_wpart+meritv03_pref_netw'

fit3_c <- cfa(model = model03,data = dat03,estimator="MLR",std.lv=FALSE) # Continuo/ estimador ML Robust

fit3_o <- cfa(model = model03,data = dat03,ordered = c("meritv03_perc_effort","meritv03_perc_talent",
                                                       "meritv03_perc_wpart","meritv03_perc_netw",
                                                       "meritv03_pref_effort","meritv03_pref_talent",
                                                       "meritv03_pref_wpart","meritv03_pref_netw"),std.lv=FALSE)
#---------------------------------------------------#
#summary(fit3_c,standardized=TRUE, fit.measures=TRUE)
fitmeasures(fit3_c)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]
semPlot::semPaths(object = fit3_c,what = "std",thresholds = FALSE, intercepts = FALSE,fade=FALSE, curve = 1.5)

#summary(fit3_o,standardized=TRUE, fit.measures=TRUE)
fitmeasures(fit3_o)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]
semPlot::semPaths(object = fit3_o,what = "std",thresholds = FALSE, intercepts = FALSE,fade=FALSE, curve = 1.5)

## chisq.scaled           df          cfi   cfi.scaled   cfi.robust 
##   57.0584003   14.0000000    0.9519967    0.9424676    0.9523832 
##        rmsea rmsea.scaled rmsea.robust 
##    0.0686923    0.0664276    0.0681436 
## chisq.scaled           df          cfi   cfi.scaled   cfi.robust 
##  60.02456981  14.00000000   0.99183602   0.98040527           NA 
##        rmsea rmsea.scaled rmsea.robust 
##   0.05074320   0.06872682           NA

NOTA: fit not very goood (RMSEA .7), no heywood. Try tau-equivalence

model03.tau <- '
perc_merit=~ r1*meritv03_perc_effort+r1*meritv03_perc_talent 
perc_nmerit=~r2*meritv03_perc_wpart+ r2*meritv03_perc_netw  
pref_merit=~ r3*meritv03_pref_effort+r3*meritv03_pref_talent 
pref_nmerit=~r4*meritv03_pref_wpart+ r4*meritv03_pref_netw'

fit3_c.tau <- cfa(model = model03.tau,data = dat03,estimator="MLR",std.lv=FALSE) # Continuo/ estimador ML Robust

fit3_o.tau <- cfa(model = model03.tau,data = dat03,ordered = c("meritv03_perc_effort","meritv03_perc_talent",
                                                               "meritv03_perc_wpart","meritv03_perc_netw",
                                                               "meritv03_pref_effort","meritv03_pref_talent",
                                                               "meritv03_pref_wpart","meritv03_pref_netw"), std.lv=FALSE)
#---------------------------------------------------#
#summary(fit3_c.tau,standardized=TRUE, fit.measures=TRUE)
fitmeasures(fit3_c.tau)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]
semPlot::semPaths(object = fit3_c.tau,what = "std",thresholds = FALSE, intercepts = FALSE,fade=FALSE, curve = 1.5)

#summary(fit3_o,standardized=TRUE, fit.measures=TRUE)
fitmeasures(fit3_o.tau)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]
semPlot::semPaths(object = fit3_o.tau,what = "std",thresholds = FALSE, intercepts = FALSE,fade=FALSE, curve = 1.5)

## chisq.scaled           df          cfi   cfi.scaled   cfi.robust 
##  54.64691498  18.00000000   0.95176935   0.95103426   0.95471247 
##        rmsea rmsea.scaled rmsea.robust 
##   0.06072423   0.05404631   0.05860871 
## chisq.scaled           df          cfi   cfi.scaled   cfi.robust 
##  59.09776813  18.00000000   0.99254921   0.98250283           NA 
##        rmsea rmsea.scaled rmsea.robust 
##   0.04275195   0.05727542           NA

rbind(fitmeasures(fit3_o)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")],
      fitmeasures(fit3_o.tau)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")])

##      chisq.scaled df       cfi cfi.scaled cfi.robust      rmsea
## [1,]     60.02457 14 0.9918360  0.9804053         NA 0.05074320
## [2,]     59.09777 18 0.9925492  0.9825028         NA 0.04275195
##      rmsea.scaled rmsea.robust
## [1,]   0.06872682           NA
## [2,]   0.05727542           NA

anova(fit3_o,fit3_o.tau)

## Scaled Chi Square Difference Test (method = "satorra.2000")
## 
##            Df AIC BIC  Chisq Chisq diff Df diff Pr(>Chisq)
## fit3_o     14         39.090                              
## fit3_o.tau 18         40.898     2.3387       4     0.6737

NOTA: El modelo con igualdad de cargas factoriales no deteriora significativamente el ajuste del modelo.

2.4 Version 04: muestra completa (N=2236)


Statistic	N	Mean	St. Dev.	Min	Pctl(25)	Pctl(75)	Max

perc_effort	2,142	3.200	1.380	1.000	2.000	4.000	5.000
perc_talent	2,141	3.024	1.161	1.000	2.000	4.000	5.000
perc_wpart	2,141	3.660	1.361	1.000	3.000	5.000	5.000
perc_netw	2,141	3.787	1.242	1.000	3.000	5.000	5.000
pref_effort	2,141	3.894	1.254	1.000	3.000	5.000	5.000
pref_talent	2,141	3.243	1.188	1.000	2.000	4.000	5.000
pref_wpart	2,141	2.688	1.175	1.000	2.000	4.000	5.000
pref_netw	2,141	2.408	1.106	1.000	2.000	3.000	5.000

cor04<-polychoric(select(dat04,perc_effort,perc_talent,perc_wpart,perc_netw,pref_effort,pref_talent,pref_wpart,pref_netw))
corrplot(cor04$rho, method="color",type = "upper", order = "hclust", tl.col = "black", tl.srt = 45,addCoef.col = "black",diag = FALSE)

model04 <- '
perc_merit =~ perc_effort+perc_talent 
perc_nmerit=~perc_wpart +perc_netw  
pref_merit =~ pref_effort+pref_talent 
pref_nmerit=~pref_wpart +pref_netw'

fit4_c <- cfa(model = model04,data = dat04,estimator="MLR",std.lv=FALSE) # Continuo/ estimador ML Robust

fit4_o <- cfa(model = model04,data = dat04,ordered = c("perc_effort","perc_talent",
                                                       "perc_wpart","perc_netw",
                                                       "pref_effort","pref_talent",
                                                       "pref_wpart","pref_netw"),std.lv=FALSE)
#---------------------------------------------------#
#summary(fit4_c,standardized=TRUE, fit.measures=TRUE)
fitmeasures(fit4_c)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]
semPlot::semPaths(object = fit3_c,what = "std",thresholds = FALSE, intercepts = FALSE,fade=FALSE,curve = 2)

#summary(fit4_o,standardized=TRUE, fit.measures=TRUE)
fitmeasures(fit4_o)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]
semPlot::semPaths(object = fit3_o,what = "std",thresholds = FALSE, intercepts = FALSE,fade=FALSE,curve = 2)

## chisq.scaled           df          cfi   cfi.scaled   cfi.robust 
## 181.10487966  14.00000000   0.95270053   0.94250163   0.95276313 
##        rmsea rmsea.scaled rmsea.robust 
##   0.07634040   0.07466588   0.07621226 
## chisq.scaled           df          cfi   cfi.scaled   cfi.robust 
## 200.77235322  14.00000000   0.99018169   0.97680332           NA 
##        rmsea rmsea.scaled rmsea.robust 
##   0.05987601   0.07895605           NA

2.4.1 Resumen ajuste ola 01

sum_fit<- bind_rows(fitmeasures(fit1_c)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")],
                    fitmeasures(fit1_o)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")],
                    fitmeasures(fit2_c)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")],
                    fitmeasures(fit2_o)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")],
                    fitmeasures(fit3_c)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")],
                    fitmeasures(fit3_o)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")],
                    fitmeasures(fit4_c)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")],
                    fitmeasures(fit4_o)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")])
sum_fit$mod <- c("Modelo 1","Modelo 1","Modelo 2","Modelo 2","Modelo 3","Modelo 3","Modelo 4","Modelo 4")
sum_fit$est <- c("Continuo","Ordinal","Continuo","Ordinal","Continuo","Ordinal","Continuo","Ordinal")
sum_fit <- select(sum_fit,mod,est,everything())

kable(sum_fit,digits = 3,format = "html",row.names = F,booktabs=T, caption = "Resumen estadísticos de ajuste ola 1") %>% kable_styling(full_width = F) %>% 
  collapse_rows(columns = 1,valign = "middle")  %>% 
  footnote(number = c("Modelo 1: fixed order by percepction/preference", 
                      "Modelo 2: fixed order by topic (i.e: effort)", 
                      "Modelo 3: Randomized order",
                      "Modelo 4: Complete sample"))

Resumen estadísticos de ajuste ola 1
mod	est	chisq.scaled	df	cfi	cfi.scaled	cfi.robust	rmsea	rmsea.scaled	rmsea.robust
Modelo 1	Continuo	42.935	14	0.979	0.977	0.980	0.058	0.054	0.057
Modelo 1	Ordinal	46.114	14	0.998	0.992		0.038	0.057
Modelo 2	Continuo	100.062	14	0.926	0.912	0.926	0.096	0.093	0.095
Modelo 2	Ordinal	101.482	14	0.984	0.962		0.071	0.094
Modelo 3	Continuo	57.058	14	0.952	0.942	0.952	0.069	0.066	0.068
Modelo 3	Ordinal	60.025	14	0.992	0.980		0.051	0.069
Modelo 4	Continuo	181.105	14	0.953	0.943	0.953	0.076	0.075	0.076
Modelo 4	Ordinal	200.772	14	0.990	0.977		0.060	0.079
¹ Modelo 1: fixed order by percepction/preference
² Modelo 2: fixed order by topic (i.e: effort)
³ Modelo 3: Randomized order
⁴ Modelo 4: Complete sample

2.5 Análisis por Grupos múltiples

model04 <- '
perc_merit =~ perc_effort+perc_talent 
perc_nmerit=~perc_wpart +perc_netw  
pref_merit =~ pref_effort+pref_talent 
pref_nmerit=~pref_wpart +pref_netw'

fit4_c_mg <- cfa(model = model04,data = dat04,estimator="MLR",std.lv=FALSE,group = "grupo") # Continuo/ estimador ML Robust

fit4_o_mg <- cfa(model = model04,data = dat04,ordered = c("perc_effort","perc_talent",
                                                       "perc_wpart","perc_netw",
                                                       "pref_effort","pref_talent",
                                                       "pref_wpart","pref_netw"),std.lv=FALSE, group = "grupo")
#---------------------------------------------------#
#summary(fit4_c_mg,standardized=TRUE, fit.measures=TRUE)
fitmeasures(fit4_c_mg)[c("cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]

##          cfi   cfi.scaled   cfi.robust        rmsea rmsea.scaled 
##   0.95494030   0.94697921   0.95532301   0.07575741   0.07314653 
## rmsea.robust 
##   0.07522301

semPlot::semPaths(object = fit4_c_mg,what = "std",thresholds = FALSE, intercepts = FALSE, fade=FALSE, curve=2)

#summary(fit4_o_mg,standardized=TRUE, fit.measures=TRUE)
fitmeasures(fit4_o_mg)[c("cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]

##          cfi   cfi.scaled   cfi.robust        rmsea rmsea.scaled 
##   0.99236975   0.98072106           NA   0.05571542   0.07542086 
## rmsea.robust 
##           NA

semPlot::semPaths(object = fit4_o_mg,what = "std",thresholds = FALSE, intercepts = FALSE,fade=FALSE, curve=2)

Cosas por probar:

V01+V02: cargas y fit
V01+V03: cargas y fit
V02+V03: cargas y fit

3 Modelo Factorial de Segundo Orden:

3.1 a. Factor de segundo orden por Percepción y Preferencia

model04b1or <- '
perc_merit=~ perc_effort+perc_talent 
perc_nmerit=~perc_wpart +perc_netw  
pref_merit=~ pref_effort+pref_talent 
pref_nmerit=~pref_wpart +pref_netw
percep=~ perc_merit + perc_nmerit
prefer=~ pref_merit + pref_nmerit'

fit4a_c <- cfa(model = model04b1or,data = dat04, estimator="MLR")

fit4b_c <- cfa(model = model04b1or,data = dat04,ordered = c("perc_effort","perc_talent","perc_wpart","perc_netw",
                                                            "pref_effort","pref_talent","pref_wpart","pref_netw"))

#---------------------------------------------------#
#summary(fit4a_c,standardized=TRUE, fit.measures=TRUE)
fitmeasures(fit4a_c)[c("cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]

##          cfi   cfi.scaled   cfi.robust        rmsea rmsea.scaled 
##   0.94736842   0.93880041   0.94760970   0.07779789   0.07441956 
## rmsea.robust 
##   0.07754042

semPlot::semPaths(object = fit4a_c,what = "std",thresholds = FALSE, intercepts = FALSE, fade=FALSE,curve=2)

#summary(fit4b_c,standardized=TRUE, fit.measures=TRUE)
fitmeasures(fit4b_c)[c("cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]

##          cfi   cfi.scaled   cfi.robust        rmsea rmsea.scaled 
##   0.98259229   0.97082290           NA   0.07702357   0.08554841 
## rmsea.robust 
##           NA

semPlot::semPaths(object = fit4b_c,what = "std",thresholds = FALSE, intercepts = FALSE, fade=FALSE,curve=2)

NOTA: Cargas factoriales de segundo orden son bajas. El ajuste del modelo no es bueno.

3.2 b. Factor de segundo orden por Meritocracia y No Meritocracia

model04b2or <-'perc_merit=~ perc_effort+perc_talent
               perc_nmerit=~perc_wpart +perc_netw
               pref_merit=~ pref_effort+pref_talent
               pref_nmerit=~pref_wpart +pref_netw
               merit =~ perc_merit  + pref_merit
               nmerit=~ perc_nmerit + pref_nmerit'

# fit4b_c <- cfa(model = model04b2or,data = dat04,estimator="MLR") # No converge

fit4b_o <- cfa(model = model04b2or,data = dat04, 
               ordered = c("perc_effort","perc_talent","perc_wpart","perc_netw",
                           "pref_effort","pref_talent","pref_wpart","pref_netw"))

#---------------------------------------------------#
# #summary(fit4b_c,standardized=TRUE, fit.measures=TRUE)
# fitmeasures(fit4b_c)[c("cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]
# semPlot::semPaths(object = fit4b_c,what = "std",thresholds = FALSE, intercepts = FALSE, fade=FALSE, curve=2)

#summary(fit4b_o,standardized=TRUE, fit.measures=TRUE)
fitmeasures(fit4b_o)[c("cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]

##          cfi   cfi.scaled   cfi.robust        rmsea rmsea.scaled 
##   0.97583700   0.95699375           NA   0.09074615   0.10386199 
## rmsea.robust 
##           NA

semPlot::semPaths(object = fit4b_o,what = "std",thresholds = FALSE, intercepts = FALSE, fade=FALSE, curve=2)

NOTA: El modelo estimado con MLR con factores de segundo orden “meritocracia/ no meritocracia” no converge. EL modelo estimado a través de DWLS converge, posee cargas muy bajas y una muy alta (merit~pref_merit). Fit es malo (RMSEA >.10)

4 Wave 02

data02 <- sjlabelled::read_spss(path = "input/data/original/Estudio_3_ola2.sav",verbose = FALSE)
load(file = "input/data/proc/datawide.RData")
dat01 <- data02 %>% select(starts_with("meritv03_p"))

4.1 ATE indicadores meritocracia

m01<- lm(perc_effort_w2~treat,data = wide01)
m02<- lm(perc_talent_w2~treat,data = wide01)
m03<- lm(perc_wpart_w2~treat, data = wide01)
m04<- lm(perc_netw_w2~treat,  data = wide01)
#--------------------------------------------#
m05<- lm(pref_effort_w2~treat,data = wide01)
m06<- lm(pref_talent_w2~treat,data = wide01)
m07<- lm(pref_wpart_w2 ~treat, data = wide01)
m08<- lm(pref_netw_w2  ~treat,  data = wide01)
#--------------------------------------------#
stargazer::stargazer(m01,m02,m03,m04,type = "html")


	Dependent variable:

	perc_effort_w2	perc_talent_w2	perc_wpart_w2	perc_netw_w2
	(1)	(2)	(3)	(4)

treatpobreza	0.061	-0.018	-0.047	-0.010
	(0.072)	(0.064)	(0.077)	(0.073)

treatdesiguald	-0.078	-0.092	0.057	0.020
	(0.072)	(0.063)	(0.077)	(0.072)

Constant	3.322^***	3.164^***	3.542^***	3.547^***
	(0.051)	(0.045)	(0.054)	(0.051)


Observations	1,777	1,777	1,777	1,777
R²	0.002	0.001	0.001	0.0001
Adjusted R²	0.001	0.0002	-0.0001	-0.001
Residual Std. Error (df = 1774)	1.239	1.094	1.326	1.249
F Statistic (df = 2; 1774)	1.844	1.175	0.907	0.084

Note:	p<0.1; p<0.05; p<0.01

stargazer::stargazer(m05,m06,m07,m08,type = "html")


	Dependent variable:

	pref_effort_w2	pref_talent_w2	pref_wpart_w2	pref_netw_w2
	(1)	(2)	(3)	(4)

treatpobreza	0.085	0.064	0.059	-0.094
	(0.067)	(0.066)	(0.067)	(0.062)

treatdesiguald	0.058	0.043	-0.017	-0.101
	(0.067)	(0.065)	(0.067)	(0.062)

Constant	3.879^***	3.237^***	2.592^***	2.374^***
	(0.047)	(0.046)	(0.047)	(0.044)


Observations	1,777	1,777	1,777	1,777
R²	0.001	0.001	0.001	0.002
Adjusted R²	-0.0002	-0.001	-0.0003	0.001
Residual Std. Error (df = 1774)	1.155	1.128	1.154	1.065
F Statistic (df = 2; 1774)	0.841	0.503	0.709	1.660

Note:	p<0.1; p<0.05; p<0.01

NOTA: Los tratamientos de información no se asocian de manera estadísticamente significativa con ninguno de los indicadores de la escala de meritocracia.

4.2 CFA Muestra completa (Control, Desigualdad y Pobreza)


Statistic	N	Mean	St. Dev.	Min	Pctl(25)	Pctl(75)	Max

meritv03_perc_effort	1,772	3.318	1.240	1.000	2.000	4.000	5.000
meritv03_perc_talent	1,772	3.125	1.095	1.000	2.000	4.000	5.000
meritv03_perc_wpart	1,772	3.547	1.325	1.000	3.000	5.000	5.000
meritv03_perc_netw	1,772	3.549	1.249	1.000	3.000	4.000	5.000
meritv03_pref_effort	1,772	3.924	1.157	1.000	3.000	5.000	5.000
meritv03_pref_talent	1,772	3.274	1.128	1.000	3.000	4.000	5.000
meritv03_pref_wpart	1,772	2.607	1.154	1.000	2.000	3.000	5.000
meritv03_pref_netw	1,772	2.308	1.065	1.000	1.000	3.000	5.000

dat01cor<- select(dat01,meritv03_perc_effort,meritv03_perc_talent,
                        meritv03_perc_wpart,meritv03_perc_netw,
                        meritv03_pref_effort,meritv03_pref_talent,
                        meritv03_pref_wpart,meritv03_pref_netw)
cor01<-polychoric(dat01cor)
corrplot(cor01$rho, type = "upper", tl.col = "black", tl.srt = 45,addCoef.col = "black",diag = FALSE)

model03 <- '
perc_merit=~meritv03_perc_effort+meritv03_perc_talent 
perc_nmerit=~meritv03_perc_wpart+meritv03_perc_netw  
pref_merit=~meritv03_pref_effort+meritv03_pref_talent 
pref_nmerit=~meritv03_pref_wpart+meritv03_pref_netw'

fitw02_c <- cfa(model = model03,data = dat01,estimator="MLR",std.lv=FALSE) # Continuo/ estimador ML Robust

fitw02_o <- cfa(model = model03,data = dat01,ordered = c("meritv03_perc_effort","meritv03_perc_talent",
                                                          "meritv03_perc_wpart","meritv03_perc_netw",
                                                          "meritv03_pref_effort","meritv03_pref_talent",
                                                          "meritv03_pref_wpart","meritv03_pref_netw"))
#---------------------------------------------------#
#summary(fitw02_c,standardized=TRUE, fit.measures=TRUE)
fitmeasures(fitw02_c)[c("cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]
semPlot::semPaths(object = fitw02_c,what = "std",thresholds = FALSE, intercepts = FALSE,fade=FALSE,curve=2)

#summary(fitw02_o,standardized=TRUE, fit.measures=TRUE)
fitmeasures(fitw02_o)[c("cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]
semPlot::semPaths(object = fitw02_o,what = "std",thresholds = FALSE, intercepts = FALSE,fade=FALSE,curve=2)

##          cfi   cfi.scaled   cfi.robust        rmsea rmsea.scaled 
##   0.96683191   0.96480728   0.96795005   0.05064198   0.04589567 
## rmsea.robust 
##   0.04967688 
##          cfi   cfi.scaled   cfi.robust        rmsea rmsea.scaled 
##   0.98784345   0.97481090           NA   0.04672993   0.05956625 
## rmsea.robust 
##           NA

NOTA: El ajuste de muestra completa en wave 02 con estimador MLR (robusto) es bastante bueno (CFI, RMSEA). El ajuste con DWLS es bastante bueno, CFI mejora, RMSEA aumenta un poco (0.59).

Recomendación: es probable que el estimador MLR sea mejor que el ordinal.

4.2.1 Factor de segundo orden

modw2 <- '
perc_merit=~meritv03_perc_effort+meritv03_perc_talent 
perc_nmerit=~meritv03_perc_wpart+meritv03_perc_netw  
pref_merit=~meritv03_pref_effort+meritv03_pref_talent 
pref_nmerit=~meritv03_pref_wpart+meritv03_pref_netw 
merit =~perc_merit+pref_merit  
nmerit=~perc_nmerit+pref_nmerit'

# mod2_merit_c <- cfa(model = modw2,data = dat01) # Continuo/ estimador ML Robust

mod2_merit_o <- cfa(model = modw2,data = dat01,ordered = c("meritv03_perc_effort","meritv03_perc_talent",
                                                           "meritv03_perc_wpart","meritv03_perc_netw",
                                                           "meritv03_pref_effort","meritv03_pref_talent",
                                                           "meritv03_pref_wpart","meritv03_pref_netw"))

#---------------------------------------------------#
# #summary(mod2_merit_c,standardized=TRUE, fit.measures=TRUE)
# fitmeasures(mod2_merit_c)[c("cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]
# semPlot::semPaths(object = mod2_merit_c,what = "std",thresholds = FALSE, intercepts = FALSE)

#summary(mod2_merit_o,standardized=TRUE, fit.measures=TRUE)
fitmeasures(mod2_merit_o)[c("cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]

##          cfi   cfi.scaled   cfi.robust        rmsea rmsea.scaled 
##   0.96479952   0.94371834           NA   0.07682146   0.08601927 
## rmsea.robust 
##           NA

semPlot::semPaths(object = mod2_merit_o,what = "std",thresholds = FALSE, intercepts = FALSE,fade=FALSE, curve=2, rotation=2)

Factor de segundo orden usando MLR no converge.

Factor de segundo orden usando DWLS converge, pero las cargas no son buenas y el ajuste tampoco.

4.3 CFA Muestra Control

w02con <- data02 %>% filter(FL_6_DO_Control==1) %>% select(starts_with("meritv03_p"))


Statistic	N	Mean	St. Dev.	Min	Pctl(25)	Pctl(75)	Max

meritv03_perc_effort	595	3.324	1.239	1.000	2.000	4.000	5.000
meritv03_perc_talent	595	3.165	1.087	1.000	2.000	4.000	5.000
meritv03_perc_wpart	595	3.536	1.296	1.000	3.000	5.000	5.000
meritv03_perc_netw	595	3.541	1.220	1.000	3.000	4.000	5.000
meritv03_pref_effort	595	3.877	1.166	1.000	3.000	5.000	5.000
meritv03_pref_talent	595	3.239	1.118	1.000	2.500	4.000	5.000
meritv03_pref_wpart	595	2.588	1.147	1.000	2.000	3.000	5.000
meritv03_pref_netw	595	2.370	1.074	1.000	2.000	3.000	5.000

corw02_1<-polychoric(w02con)
corrplot(corw02_1$rho, type = "upper", order = "hclust", tl.col = "black", tl.srt = 45,addCoef.col = "black",diag = FALSE)

model03 <- '
perc_merit=~meritv03_perc_effort+meritv03_perc_talent 
perc_nmerit=~meritv03_perc_wpart+meritv03_perc_netw  
pref_merit=~meritv03_pref_effort+meritv03_pref_talent 
pref_nmerit=~meritv03_pref_wpart+meritv03_pref_netw'

fitw02a_c <- cfa(model = model03,data = w02con,estimator="MLR") # Continuo/ estimador ML Robust

fitw02a_o <- cfa(model = model03,data = w02con,ordered = c("meritv03_perc_effort","meritv03_perc_talent",
                                                           "meritv03_perc_wpart","meritv03_perc_netw",
                                                           "meritv03_pref_effort","meritv03_pref_talent",
                                                           "meritv03_pref_wpart","meritv03_pref_netw"))

#---------------------------------------------------#
#summary(fitw02a_c,standardized=TRUE, fit.measures=TRUE)
fitmeasures(fitw02a_c)[c("cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]
semPlot::semPaths(object = fitw02a_c,what = "std",thresholds = FALSE, intercepts = FALSE, curve=2)

#summary(fitw02a_o,standardized=TRUE, fit.measures=TRUE)
fitmeasures(fitw02a_o)[c("cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]
semPlot::semPaths(object = fitw02a_o,what = "std",thresholds = FALSE, intercepts = FALSE, curve=2)

##          cfi   cfi.scaled   cfi.robust        rmsea rmsea.scaled 
##   0.95300796   0.95202309   0.95683934   0.05713915   0.04989786 
## rmsea.robust 
##   0.05433717 
##          cfi   cfi.scaled   cfi.robust        rmsea rmsea.scaled 
##   0.98848464   0.97507075           NA   0.04333340   0.05692432 
## rmsea.robust 
##           NA

4.4 CFA Muestra Pobreza

w02pob <- data02 %>% filter(FL_6_DO_Tratamiento==1) %>% select(starts_with("meritv03_p"))


Statistic	N	Mean	St. Dev.	Min	Pctl(25)	Pctl(75)	Max

meritv03_perc_effort	586	3.387	1.260	1.000	2.000	4.000	5.000
meritv03_perc_talent	586	3.140	1.099	1.000	2.000	4.000	5.000
meritv03_perc_wpart	586	3.493	1.368	1.000	3.000	5.000	5.000
meritv03_perc_netw	586	3.534	1.274	1.000	3.000	4.000	5.000
meritv03_pref_effort	586	3.956	1.173	1.000	4.000	5.000	5.000
meritv03_pref_talent	586	3.304	1.135	1.000	3.000	4.000	5.000
meritv03_pref_wpart	586	2.654	1.142	1.000	2.000	3.000	5.000
meritv03_pref_netw	586	2.280	1.085	1.000	1.000	3.000	5.000

corw02_2<-polychoric(w02pob)
corrplot(corw02_2$rho, type = "upper", order = "hclust", tl.col = "black", tl.srt = 45,addCoef.col = "black",diag = FALSE)

model03 <- '
perc_merit=~meritv03_perc_effort+meritv03_perc_talent 
perc_nmerit=~meritv03_perc_wpart+meritv03_perc_netw  
pref_merit=~meritv03_pref_effort+meritv03_pref_talent 
pref_nmerit=~meritv03_pref_wpart+meritv03_pref_netw'

fitw02b_c <- cfa(model = model03,data = w02pob,estimator="MLR") # Continuo/ estimador ML Robust

fitw02b_o <- cfa(model = model03,data = w02pob,ordered = c("meritv03_perc_effort","meritv03_perc_talent",
                                                     "meritv03_perc_wpart","meritv03_perc_netw",
                                                     "meritv03_pref_effort","meritv03_pref_talent",
                                                     "meritv03_pref_wpart","meritv03_pref_netw"))

#---------------------------------------------------#
#summary(fitw02b_c,standardized=TRUE, fit.measures=TRUE)
fitmeasures(fitw02b_c)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]
semPlot::semPaths(object = fitw02b_c,what = "std",thresholds = FALSE, intercepts = FALSE, curve=2)

#summary(fitw02b_o,standardized=TRUE, fit.measures=TRUE)
fitmeasures(fitw02b_o)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]
semPlot::semPaths(object = fitw02b_o,what = "std",thresholds = FALSE, intercepts = FALSE, curve=2)

## chisq.scaled           df          cfi   cfi.scaled   cfi.robust 
##  21.89512187  14.00000000   0.98686103   0.98728536   0.98891906 
##        rmsea rmsea.scaled rmsea.robust 
##   0.03601966   0.03102177   0.03291156 
## chisq.scaled           df          cfi   cfi.scaled   cfi.robust 
##  32.82736205  14.00000000   0.99739823   0.98732355           NA 
##        rmsea rmsea.scaled rmsea.robust 
##   0.02531526   0.04794604           NA

4.5 CFA Muestra Desigualdad

w02des <- data02 %>% filter(FL_6_DO_Tratamiento_desigualdad==1) %>% select(starts_with("meritv03_p"))


Statistic	N	Mean	St. Dev.	Min	Pctl(25)	Pctl(75)	Max

meritv03_perc_effort	591	3.244	1.219	1.000	2.000	4.000	5.000
meritv03_perc_talent	591	3.069	1.099	1.000	2.000	4.000	5.000
meritv03_perc_wpart	591	3.611	1.311	1.000	3.000	5.000	5.000
meritv03_perc_netw	591	3.572	1.255	1.000	3.000	4.000	5.000
meritv03_pref_effort	591	3.939	1.130	1.000	3.000	5.000	5.000
meritv03_pref_talent	591	3.281	1.132	1.000	3.000	4.000	5.000
meritv03_pref_wpart	591	2.579	1.173	1.000	2.000	3.000	5.000
meritv03_pref_netw	591	2.272	1.034	1.000	1.000	3.000	5.000

corw02_3<-polychoric(w02des)
corrplot(corw02_3$rho, type = "upper", order = "hclust", tl.col = "black", tl.srt = 45,addCoef.col = "black",diag = FALSE)

model03 <- '
perc_merit=~meritv03_perc_effort+meritv03_perc_talent
perc_nmerit=~meritv03_perc_wpart+meritv03_perc_netw  
pref_merit=~meritv03_pref_effort+meritv03_pref_talent 
pref_nmerit=~meritv03_pref_wpart+meritv03_pref_netw'

fitw02c_c <- cfa(model = model03,data = w02des,estimator="MLR") # Continuo/ estimador ML Robust

fitw02c_o <- cfa(model = model03,data = w02des,ordered = c("meritv03_perc_effort","meritv03_perc_talent",
                                                           "meritv03_perc_wpart","meritv03_perc_netw",
                                                           "meritv03_pref_effort","meritv03_pref_talent",
                                                           "meritv03_pref_wpart","meritv03_pref_netw"))
#---------------------------------------------------#
#summary(fitw02c_c,standardized=TRUE, fit.measures=TRUE)
fitmeasures(fitw02c_c)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]
semPlot::semPaths(object = fitw02c_c,what = "std",thresholds = FALSE, intercepts = FALSE)

#summary(fitw02b_o,standardized=TRUE, fit.measures=TRUE)
fitmeasures(fitw02c_o)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]
semPlot::semPaths(object = fitw02c_o,what = "std",thresholds = FALSE, intercepts = FALSE)

## chisq.scaled           df          cfi   cfi.scaled   cfi.robust 
##  46.32268992  14.00000000   0.93940521   0.92687174   0.93936161 
##        rmsea rmsea.scaled rmsea.robust 
##   0.06371050   0.06250230   0.06334778 
## chisq.scaled           df          cfi   cfi.scaled   cfi.robust 
##  69.67269301  14.00000000   0.97297449   0.94366656           NA 
##        rmsea rmsea.scaled rmsea.robust 
##   0.06420276   0.08209772           NA

5 Resumen ajuste ola 2

sum_fitw02<- bind_rows(fitmeasures(fitw02_c)[ c("cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")],
                       fitmeasures(fitw02_o)[ c("cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")],
                       fitmeasures(fitw02a_c)[c("cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")],
                       fitmeasures(fitw02a_o)[c("cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")],
                       fitmeasures(fitw02b_c)[c("cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")],
                       fitmeasures(fitw02b_o)[c("cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")],
                       fitmeasures(fitw02c_c)[c("cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")],
                       fitmeasures(fitw02c_c)[c("cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")])

sum_fitw02$mod <- c("Modelo 1","Modelo 1","Modelo 2","Modelo 2","Modelo 3","Modelo 3","Modelo 4","Modelo 4")
sum_fitw02$est <- c("Continuo","Ordinal","Continuo","Ordinal","Continuo","Ordinal","Continuo","Ordinal")
sum_fitw02 <- select(sum_fitw02,mod,est,everything())

kable(sum_fitw02,digits = 3,format = "html",row.names = F,booktabs=T, caption = "Resumen ajuste ola 2") %>% 
  kable_styling(full_width = F) %>% 
  collapse_rows(columns = 1,valign = "middle") %>% 
  footnote(number = c("Modelo 1: Muestra completa", 
                      "Modelo 2: Muestra grupo Control", 
                      "Modelo 3: Muestra grupo tratamiento pobreza",
                      "Modelo 4: Muestra grupo tratamiento desigualdad"))

Resumen ajuste ola 2
mod	est	cfi	cfi.scaled	cfi.robust	rmsea	rmsea.scaled	rmsea.robust
Modelo 1	Continuo	0.967	0.965	0.968	0.051	0.046	0.050
Modelo 1	Ordinal	0.988	0.975		0.047	0.060
Modelo 2	Continuo	0.953	0.952	0.957	0.057	0.050	0.054
Modelo 2	Ordinal	0.988	0.975		0.043	0.057
Modelo 3	Continuo	0.987	0.987	0.989	0.036	0.031	0.033
Modelo 3	Ordinal	0.997	0.987		0.025	0.048
Modelo 4	Continuo	0.939	0.927	0.939	0.064	0.063	0.063
Modelo 4	Ordinal	0.939	0.927	0.939	0.064	0.063	0.063
¹ Modelo 1: Muestra completa
² Modelo 2: Muestra grupo Control
³ Modelo 3: Muestra grupo tratamiento pobreza
⁴ Modelo 4: Muestra grupo tratamiento desigualdad

6 Wave 03

data03 <- sjlabelled::read_spss(path = "input/data/original/Estudio_3_ola3.sav",verbose = FALSE)
dat01 <- data03 %>% select(starts_with("meritv03_p"))

sjt.fa(data = dat01,show.comm = TRUE)

## Parallel analysis suggests that the number of factors =  3  and the number of components =  NA

Factor Analysis
	Factor 1	Factor 2	Factor 3	Communality
Pensando en la sociedad chilena, ¿En qué medida se encuentra usted de acuerdo o en desacuerdo con cada una de las siguientes afirmaciones? - Quienes más se esfuerzan logran obtener mayores recompensas que quienes se esfuerzan menos.	0.66	-0.28	0.09	0.45
Pensando en la sociedad chilena, ¿En qué medida se encuentra usted de acuerdo o en desacuerdo con cada una de las siguientes afirmaciones? - Quienes poseen más talento logran obtener mayores recompensas que quienes poseen menos talento.	0.58	-0.09	0.13	0.36
Pensando en la sociedad chilena, ¿En qué medida se encuentra usted de acuerdo o en desacuerdo con cada una de las siguientes afirmaciones? - Quienes tienen padres ricos logran salir adelante.	-0.05	0.73	0.08	0.52
Pensando en la sociedad chilena, ¿En qué medida se encuentra usted de acuerdo o en desacuerdo con cada una de las siguientes afirmaciones? - Quienes tienen buenos contactos logran salir adelante.	-0.02	0.71	0.09	0.50
Pensando en la sociedad chilena, ¿En qué medida se encuentra usted de acuerdo o en desacuerdo con cada una de las siguientes afirmaciones? - Quienes más se esfuerzan deberían obtener mayores recompensas que quienes se esfuerzan menos.	0.55	0.19	-0.15	0.39
Pensando en la sociedad chilena, ¿En qué medida se encuentra usted de acuerdo o en desacuerdo con cada una de las siguientes afirmaciones? - Quienes poseen más talento deberían obtener mayores recompensas que quienes poseen menos talento.	0.43	0.23	-0.07	0.28
Pensando en la sociedad chilena, ¿En qué medida se encuentra usted de acuerdo o en desacuerdo con cada una de las siguientes afirmaciones? - Está bien que quienes tienen padres ricos salgan adelante.	0.03	0.13	0.75	0.58
Pensando en la sociedad chilena, ¿En qué medida se encuentra usted de acuerdo o en desacuerdo con cada una de las siguientes afirmaciones? - Está bien que quienes tienen buenos contactos salgan adelante.	0.02	0.06	0.69	0.49
Total Communalities				3.58
Cronbach’s α	0.62	0.68	0.69

model03 <- '
perc_merit=~meritv03_perc_effort+meritv03_perc_talent 
perc_nmerit=~meritv03_perc_wpart+meritv03_perc_netw  
pref_merit=~meritv03_pref_effort+meritv03_pref_talent 
pref_nmerit=~meritv03_pref_wpart+meritv03_pref_netw'

fitw03c_c <- cfa(model = model03,data = dat01,estimator="MLR") # Continuo/ estimador ML Robust

fitw03c_o <- cfa(model = model03,data = dat01,ordered = c("meritv03_perc_effort","meritv03_perc_talent",
                                                          "meritv03_perc_wpart","meritv03_perc_netw",
                                                          "meritv03_pref_effort","meritv03_pref_talent",
                                                          "meritv03_pref_wpart","meritv03_pref_netw"))
#---------------------------------------------------#
#summary(fitw03c_c,standardized=TRUE, fit.measures=TRUE)
fitmeasures(fitw03c_c)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]

## chisq.scaled           df          cfi   cfi.scaled   cfi.robust 
##  62.42098851  14.00000000   0.97001271   0.96743840   0.97104967 
##        rmsea rmsea.scaled rmsea.robust 
##   0.05354429   0.04868836   0.05249427

#summary(fitw03c_o,standardized=TRUE, fit.measures=TRUE)
fitmeasures(fitw03c_o)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]

## chisq.scaled           df          cfi   cfi.scaled   cfi.robust 
## 106.11132208  14.00000000   0.98842335   0.97425449           NA 
##        rmsea rmsea.scaled rmsea.robust 
##   0.05216642   0.06717594           NA

summary > * Ajuste usando estimador MLR es bueno (CFI = 0.971, RMSEA = 0.052). El chi2 (robusto) es 62.421
> * Ajuste usando estimador DWLS es bueno (CFI = 0.974, RMSEA = 0.067). El chi2 (robusto) es 106.111

7 Resumen ajuste ola 3

sum_fitw03<- bind_rows(fitmeasures(fitw03c_c)[ c("cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")],
                       fitmeasures(fitw03c_o)[ c("cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")])

sum_fitw03$mod <- c("Modelo 1","Modelo 1")
sum_fitw03$est <- c("Continuo","Ordinal")
sum_fitw03 <- select(sum_fitw03,mod,est,everything())

kable(sum_fitw03,digits = 3,format = "html",row.names = F,booktabs=T,caption = "Resumen ajuste ola 3") %>% 
  kable_styling(full_width = F) %>% 
  collapse_rows(columns = 1,valign = "middle") %>% 
  footnote(number = c("Modelo 1: Muestra completa"))

Resumen ajuste ola 3
mod	est	cfi	cfi.scaled	cfi.robust	rmsea	rmsea.scaled	rmsea.robust
Modelo 1	Continuo	0.970	0.967	0.971	0.054	0.049	0.052
Modelo 1	Ordinal	0.988	0.974		0.052	0.067
¹ Modelo 1: Muestra completa

8 Análisis longitudinal

library(sjPlot)
library(dplyr)
library(lavaan)
library(stargazer)
library(corrplot)
library(psych)
library(knitr)
library(kableExtra)
load(file = "input/data/proc/datawide_123.RData")

8.1 Correlación indicadores w01 y w02

g01<- wide02 %>% filter(grupo==1) %>% select(starts_with("perc_"),starts_with("pref_")) # Fixed by perception/preference
g02<- wide02 %>% filter(grupo==2) %>% select(starts_with("perc_"),starts_with("pref_")) # Fixed by topic (i.e: effort)
g03<- wide02 %>% filter(grupo==3) %>% select(starts_with("perc_"),starts_with("pref_")) # Randomized

8.1.1 Grupo Versión 01

stargazer(g01,type = "html", median = T,digits = 2)


Statistic	N	Mean	St. Dev.	Min	Pctl(25)	Median	Pctl(75)	Max

perc_effort_w1	486	3.22	1.35	1	2	3	4	5
perc_talent_w1	486	3.02	1.11	1	2	3	4	5
perc_wpart_w1	486	3.61	1.36	1	3	4	5	5
perc_netw_w1	486	3.76	1.23	1	3	4	5	5
perc_effort_w2	486	3.33	1.23	1	2	3.5	4	5
perc_talent_w2	486	3.13	1.09	1	2	3	4	5
perc_wpart_w2	486	3.59	1.30	1	3	4	5	5
perc_netw_w2	486	3.56	1.26	1	3	4	4	5
perc_effort_w3	486	3.47	1.16	1	3	4	4	5
perc_talent_w3	486	3.18	1.01	1	3	3	4	5
perc_wpart_w3	486	3.38	1.30	1	3	4	4	5
perc_netw_w3	486	3.52	1.20	1	3	4	4	5
pref_effort_w1	486	3.91	1.24	1	3	4	5	5
pref_talent_w1	486	3.33	1.15	1	3	3	4	5
pref_wpart_w1	486	2.54	1.12	1	2	3	3	5
pref_netw_w1	486	2.30	1.06	1	1	2	3	5
pref_effort_w2	486	3.92	1.13	1	3	4	5	5
pref_talent_w2	486	3.30	1.10	1	3	3	4	5
pref_wpart_w2	486	2.60	1.14	1	2	3	3	5
pref_netw_w2	486	2.29	1.09	1	1	2	3	5
pref_effort_w3	486	4.03	1.11	1	4	4	5	5
pref_talent_w3	486	3.49	1.01	1	3	4	4	5
pref_wpart_w3	486	2.51	1.16	1	2	3	3	5
pref_netw_w3	486	2.35	1.08	1	2	2	3	5

corwide01a<- polychoric(x = select(g01 ,starts_with(match = "perc_")),na.rm = T)
corrplot(corwide01a$rho, method = "color" ,type = "upper", tl.col = "black", tl.srt = 45,addCoef.col = "black",diag = FALSE)

Percepción

corwide01b<- polychoric(x = select(g01 ,starts_with(match = "pref_")),na.rm = T)
corrplot(corwide01b$rho, method = "color" ,type = "upper", tl.col = "black", tl.srt = 45,addCoef.col = "black",diag = FALSE)

Preferencia

8.1.2 Grupo Versión 02

stargazer(g02,type = "html", median = T,digits = 2)


Statistic	N	Mean	St. Dev.	Min	Pctl(25)	Median	Pctl(75)	Max

perc_effort_w1	471	3.10	1.42	1	2	3	4	5
perc_talent_w1	471	3.09	1.13	1	2	3	4	5
perc_wpart_w1	471	3.65	1.36	1	3	4	5	5
perc_netw_w1	471	3.68	1.22	1	3	4	5	5
perc_effort_w2	471	3.35	1.25	1	2	4	4	5
perc_talent_w2	471	3.14	1.10	1	2	3	4	5
perc_wpart_w2	471	3.59	1.26	1	3	4	5	5
perc_netw_w2	471	3.61	1.20	1	3	4	4	5
perc_effort_w3	471	3.54	1.22	1	3	4	5	5
perc_talent_w3	471	3.21	1.06	1	2.5	3	4	5
perc_wpart_w3	471	3.41	1.27	1	3	4	4	5
perc_netw_w3	471	3.43	1.28	1	3	4	4	5
pref_effort_w1	471	3.75	1.34	1	3	4	5	5
pref_talent_w1	471	3.18	1.15	1	2	3	4	5
pref_wpart_w1	471	2.77	1.20	1	2	3	4	5
pref_netw_w1	471	2.39	1.09	1	2	2	3	5
pref_effort_w2	471	4.00	1.14	1	4	4	5	5
pref_talent_w2	471	3.27	1.15	1	2.5	3	4	5
pref_wpart_w2	471	2.59	1.15	1	2	3	3	5
pref_netw_w2	471	2.30	1.03	1	2	2	3	5
pref_effort_w3	471	4.11	1.08	1	4	4	5	5
pref_talent_w3	471	3.45	1.12	1	3	4	4	5
pref_wpart_w3	471	2.53	1.14	1	2	3	3	5
pref_netw_w3	471	2.28	1.05	1	1	2	3	5

corwide02a<- polychoric(x = select(g02 ,starts_with(match = "perc_")),na.rm = T)
corrplot(corwide02a$rho, method = "color" ,type = "upper", tl.col = "black", tl.srt = 45,addCoef.col = "black",diag = FALSE)

Percepción

corwide02b<- polychoric(x = select(g02 ,starts_with(match = "pref_")),na.rm = T)
corrplot(corwide02b$rho, method = "color" ,type = "upper", tl.col = "black", tl.srt = 45,addCoef.col = "black",diag = FALSE)

Preferencia

8.1.3 Grupo Versión 03

stargazer(g03,type = "html", median = T,digits = 2)


Statistic	N	Mean	St. Dev.	Min	Pctl(25)	Median	Pctl(75)	Max

perc_effort_w1	491	3.41	1.34	1	2	4	5	5
perc_talent_w1	491	3.12	1.24	1	2	3	4	5
perc_wpart_w1	491	3.76	1.34	1	3	4	5	5
perc_netw_w1	491	3.92	1.22	1	3	4	5	5
perc_effort_w2	491	3.32	1.24	1	2	4	4	5
perc_talent_w2	491	3.13	1.09	1	2	3	4	5
perc_wpart_w2	491	3.44	1.41	1	2	4	5	5
perc_netw_w2	491	3.50	1.28	1	3	4	4	5
perc_effort_w3	491	3.49	1.23	1	3	4	4.5	5
perc_talent_w3	491	3.21	1.01	1	3	3	4	5
perc_wpart_w3	491	3.41	1.31	1	3	4	4	5
perc_netw_w3	491	3.40	1.25	1	3	4	4	5
pref_effort_w1	491	4.14	1.11	1	4	4	5	5
pref_talent_w1	491	3.41	1.24	1	3	4	4	5
pref_wpart_w1	491	2.79	1.19	1	2	3	4	5
pref_netw_w1	491	2.48	1.15	1	2	2	3	5
pref_effort_w2	491	3.88	1.20	1	3	4	5	5
pref_talent_w2	491	3.30	1.10	1	3	3	4	5
pref_wpart_w2	491	2.58	1.16	1	2	3	3	5
pref_netw_w2	491	2.32	1.04	1	2	2	3	5
pref_effort_w3	491	4.05	1.08	1	4	4	5	5
pref_talent_w3	491	3.45	1.09	1	3	3	4	5
pref_wpart_w3	491	2.53	1.14	1	2	3	3	5
pref_netw_w3	491	2.34	1.07	1	1	2	3	5

corwide03a<- polychoric(x = select(g03 ,starts_with(match = "perc_")),na.rm = T)
corrplot(corwide03a$rho, type = "upper", tl.col = "black", tl.srt = 45,addCoef.col = "black",diag = FALSE,tl.cex = 0.75)

Percepción

corwide03b<- polychoric(x = select(g03 ,starts_with(match = "pref_")),na.rm = T)
corrplot(corwide03b$rho, method = "color" ,type = "upper", tl.col = "black", tl.srt = 45,addCoef.col = "black",diag = FALSE)

Preferencia

8.1.4 Muestra completa

stargazer(select(wide02 ,starts_with(match = "perc_"), starts_with("pref_")),type = "html", median = T,digits = 2)


Statistic	N	Mean	St. Dev.	Min	Pctl(25)	Median	Pctl(75)	Max

perc_effort_w1	1,448	3.24	1.38	1	2	3	4	5
perc_talent_w1	1,448	3.08	1.16	1	2	3	4	5
perc_wpart_w1	1,448	3.67	1.35	1	3	4	5	5
perc_netw_w1	1,448	3.79	1.23	1	3	4	5	5
perc_effort_w2	1,448	3.33	1.24	1	2	4	4	5
perc_talent_w2	1,448	3.13	1.09	1	2	3	4	5
perc_wpart_w2	1,448	3.54	1.33	1	3	4	5	5
perc_netw_w2	1,448	3.55	1.25	1	3	4	4	5
perc_effort_w3	1,448	3.50	1.20	1	3	4	4	5
perc_talent_w3	1,448	3.20	1.03	1	3	3	4	5
perc_wpart_w3	1,448	3.40	1.29	1	3	4	4	5
perc_netw_w3	1,448	3.45	1.24	1	3	4	4	5
pref_effort_w1	1,448	3.93	1.24	1	3	4	5	5
pref_talent_w1	1,448	3.31	1.18	1	3	3	4	5
pref_wpart_w1	1,448	2.70	1.18	1	2	3	4	5
pref_netw_w1	1,448	2.39	1.10	1	1.8	2	3	5
pref_effort_w2	1,448	3.93	1.16	1	3	4	5	5
pref_talent_w2	1,448	3.29	1.12	1	3	3	4	5
pref_wpart_w2	1,448	2.59	1.15	1	2	3	3	5
pref_netw_w2	1,448	2.30	1.05	1	1	2	3	5
pref_effort_w3	1,448	4.06	1.09	1	4	4	5	5
pref_talent_w3	1,448	3.46	1.07	1	3	4	4	5
pref_wpart_w3	1,448	2.53	1.15	1	2	3	3	5
pref_netw_w3	1,448	2.32	1.07	1	1	2	3	5

corwideA<- polychoric(x = select(wide02 ,starts_with(match = "perc_")),na.rm = T)
corrplot(corwideA$rho, method = "color" ,type = "upper", tl.col = "black", tl.srt = 45,addCoef.col = "black",diag = FALSE)

Percepción

corwideB<- polychoric(x = select(wide02 ,starts_with(match = "pref_")),na.rm = T)
corrplot(corwideB$rho, method = "color" ,type = "upper", tl.col = "black", tl.srt = 45,addCoef.col = "black",diag = FALSE)

Preferencia

8.2 CFA wave01, wave02 y wave03

model_long <- '
# wave 01
perc_merit1 =~perc_effort_w1+perc_talent_w1 
perc_nmerit1=~perc_wpart_w1 +perc_netw_w1  
pref_merit1 =~pref_effort_w1+pref_talent_w1 
pref_nmerit1=~pref_wpart_w1 +pref_netw_w1
# wave 02
perc_merit2 =~perc_effort_w2+perc_talent_w2 
perc_nmerit2=~perc_wpart_w2 +perc_netw_w2  
pref_merit2 =~pref_effort_w2+pref_talent_w2 
pref_nmerit2=~pref_wpart_w2 +pref_netw_w2
# wave 03
perc_merit3 =~perc_effort_w3+perc_talent_w3 
perc_nmerit3=~perc_wpart_w3 +perc_netw_w3  
pref_merit3 =~pref_effort_w3+pref_talent_w3 
pref_nmerit3=~pref_wpart_w3 +pref_netw_w3'


fitlong_c <- cfa(model = model_long,data = wide02,estimator="MLR") # Continuo/ estimador ML Robust

indicadores <-c("perc_effort_w1","perc_talent_w1","perc_wpart_w1","perc_netw_w1",
                "pref_effort_w1","pref_talent_w1","pref_wpart_w1","pref_netw_w1",
                "perc_effort_w2","perc_talent_w2","perc_wpart_w2","perc_netw_w2",
                "pref_effort_w2","pref_talent_w2","pref_wpart_w2","pref_netw_w2")  

fitlong_o <- cfa(model = model_long,data = wide02,ordered = indicadores)

#---------------------------------------------------#
#summary(fitlong_c,standardized=TRUE, fit.measures=TRUE)
fitmeasures(fitlong_c)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]

## chisq.scaled           df          cfi   cfi.scaled   cfi.robust 
## 665.88017314 186.00000000   0.93012912   0.92988159   0.93302455 
##        rmsea rmsea.scaled rmsea.robust 
##   0.04637638   0.04221100   0.04525147

semPlot::semPaths(object = fitlong_c,what = "std",thresholds = FALSE, intercepts = FALSE)

#summary(fitlong_o,standardized=TRUE, fit.measures=TRUE)
fitmeasures(fitlong_o)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]

## chisq.scaled           df          cfi   cfi.scaled   cfi.robust 
## 747.48098303 186.00000000   0.98273020   0.94727999           NA 
##        rmsea rmsea.scaled rmsea.robust 
##   0.03597639   0.04567481           NA

semPlot::semPaths(object = fitlong_o,what = "std",thresholds = FALSE, intercepts = FALSE,style = "mx")

cormat<- as.matrix(inspect(fitlong_o, "cor.lv"))
corrplot(cormat, method = "color" ,type = "upper", tl.col = "black", tl.srt = 45,addCoef.col = "black",diag = FALSE)

Correlación entre variables latentes w01, w02 y w03

9 Invarianza

Invarianza entre modalidades en ola 1,2 y 3 (between groups invariance)
Invarianza entre tratamientos en ola 2(between groups invariance)

NOTA: En vista de que no existe una librería que permita estimar invarianza longitudinal asumiendo indicadores ordinales, las primeras estimaciones las realizaré asumiendo indicadores continuos (estimator=ML).

library(semTools)

load(file = "input/data/proc/datalong.RData")     # Datos formato long
load(file = "input/data/proc/datawide_123.RData") # Datos formato long

dim(dat04)

## [1] 2236   36

m_inv_modalidad <- 'perc_merit=~ perc_effort+perc_talent 
                    perc_nmerit=~perc_wpart +perc_netw  
                    pref_merit=~ pref_effort+pref_talent 
                    pref_nmerit=~pref_wpart +pref_netw'

indicators  <-c("perc_effort","perc_talent","perc_wpart","perc_netw",
                "pref_effort","pref_talent","pref_wpart","pref_netw")  

fit.conf=cfa(m_inv_modalidad,data=dat04,group="grupo")
fit.deb =cfa(m_inv_modalidad,data=dat04,group="grupo",group.equal=c("loadings"))
fit.fuer=cfa(m_inv_modalidad,data=dat04,group="grupo",group.equal=c("loadings","intercepts"))
fit.str =cfa(m_inv_modalidad,data=dat04,group="grupo",group.equal=c("loadings","intercepts","residuals"))


fitmeasures(fit.conf)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]

##        <NA>          df         cfi        <NA>        <NA>       rmsea 
##          NA 42.00000000  0.95494030          NA          NA  0.07575741 
##        <NA>        <NA> 
##          NA          NA

fitmeasures(fit.deb )[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]

##        <NA>          df         cfi        <NA>        <NA>       rmsea 
##          NA 50.00000000  0.94850216          NA          NA  0.07422755 
##        <NA>        <NA> 
##          NA          NA

fitmeasures(fit.fuer)[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]

##        <NA>          df         cfi        <NA>        <NA>       rmsea 
##          NA 58.00000000  0.94733603          NA          NA  0.06969449 
##        <NA>        <NA> 
##          NA          NA

fitmeasures(fit.str )[c("chisq.scaled","df","cfi","cfi.scaled","cfi.robust","rmsea","rmsea.scaled","rmsea.robust")]

##        <NA>          df         cfi        <NA>        <NA>       rmsea 
##          NA 74.00000000  0.91887795          NA          NA  0.07657891 
##        <NA>        <NA> 
##          NA          NA

anova(fit.conf, fit.deb, fit.fuer, fit.str)

## Chi Square Difference Test
## 
##          Df   AIC   BIC  Chisq Chisq diff Df diff            Pr(>Chisq)
## fit.conf 42 52068 52578 214.03                                         
## fit.deb  50 52084 52549 246.61     32.579       8            0.00007332
## fit.fuer 58 52081 52500 259.06     12.452       8                0.1321
## fit.str  74 52173 52502 383.70    124.646      16 < 0.00000000000000022
##             
## fit.conf    
## fit.deb  ***
## fit.fuer    
## fit.str  ***
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1

configural <- '
# wave 01
perc_merit1 =~NA*perc_effort_w1+lambda1*perc_effort_w1+perc_talent_w1 
perc_nmerit1=~NA*perc_wpart_w1 +lambda1*perc_wpart_w1 +perc_netw_w1  
pref_merit1 =~NA*pref_effort_w1+lambda1*pref_effort_w1+pref_talent_w1 
pref_nmerit1=~NA*pref_wpart_w1 +lambda1*pref_wpart_w1 +pref_netw_w1
# wave 02
perc_merit2 =~NA*perc_effort_w2+lambda1*perc_effort_w2+perc_talent_w2 
perc_nmerit2=~NA*perc_wpart_w2 +lambda1*perc_wpart_w2 +perc_netw_w2  
pref_merit2 =~NA*pref_effort_w2+lambda1*pref_effort_w2+pref_talent_w2 
pref_nmerit2=~NA*pref_wpart_w2 +lambda1*pref_wpart_w2 +pref_netw_w2
# wave 03
perc_merit3 =~NA*perc_effort_w3+lambda1*perc_effort_w3+perc_talent_w3 
perc_nmerit3=~NA*perc_wpart_w3 +lambda1*perc_wpart_w3 +perc_netw_w3  
pref_merit3 =~NA*pref_effort_w3+lambda1*pref_effort_w3+pref_talent_w3 
pref_nmerit3=~NA*pref_wpart_w3 +lambda1*pref_wpart_w3 +pref_netw_w3

# INTERCEPTS: Por cada variable latente, debo fijar el intercepto del primer indicador.
# wave 01--------------------------
perc_effort_w1 ~ i1*1   # Percepcion merito        
perc_talent_w1 ~ 1      #      
perc_wpart_w1 ~ i1*1    # Percepcion no-merito   
perc_netw_w1 ~ 1        #
pref_effort_w1 ~ i1*1   # Preferencia merito    
pref_talent_w1 ~ 1      #  
pref_wpart_w1 ~ i1*1    # Preferencia no-merito   
pref_netw_w1 ~ 1        #

# wave 02--------------------------
perc_effort_w2 ~ i1*1   # Percepcion merito                    
perc_talent_w2 ~ 1      #               
perc_wpart_w2 ~ i1*1    # Percepcion no-merito              
perc_netw_w2 ~ 1        #       
pref_effort_w2 ~ i1*1   # Preferencia merito                
pref_talent_w2 ~ 1      #           
pref_wpart_w2 ~ i1*1    # Preferencia no-merito              
pref_netw_w2 ~ 1        #       
               
# wave 03--------------------------
perc_effort_w3 ~ i1*1   # Percepcion merito                     
perc_talent_w3 ~ 1      #                            
perc_wpart_w3 ~ i1*1    # Percepcion no-merito                
perc_netw_w3 ~ 1        #                  
pref_effort_w3 ~ i1*1   # Preferencia merito                   
pref_talent_w3 ~ 1      #                        
pref_wpart_w3 ~ i1*1    # Preferencia no-merito               
pref_netw_w3 ~ 1        #

# UNIQUE VARIANCES and COVARIANCES.


# LATENT VARIABLES MEANS: Para cada variable latente fijar la media
# Fix T1 common factor mean to zero
perc_merit1 ~ 0*1
# Freely estimate T2-T3 common factor means
perc_merit2 ~ 1
perc_merit3 ~ 1

perc_nmerit1 ~ 0*1
perc_nmerit2 ~ 1
perc_nmerit3 ~ 1

pref_merit1 ~ 0*1
pref_merit2 ~ 1
pref_merit3 ~ 1

pref_nmerit1 ~ 0*1
pref_nmerit2 ~ 1
pref_nmerit3 ~ 1

# LATENT VARIABLE VARIANCES AND COVARIANCES: correlaciones entre variables latentes

'                   

fit_conf <- cfa(model = configural,data =wide02)
summary(fit_conf, standardized=TRUE)

## lavaan 0.6-3 ended normally after 87 iterations
## 
##   Optimization method                           NLMINB
##   Number of free parameters                        158
##   Number of equality constraints                    22
## 
##   Number of observations                          1448
## 
##   Estimator                                         ML
##   Model Fit Test Statistic                    1460.610
##   Degrees of freedom                               188
##   P-value (Chi-square)                           0.000
## 
## Parameter Estimates:
## 
##   Information                                 Expected
##   Information saturated (h1) model          Structured
##   Standard Errors                             Standard
## 
## Latent Variables:
##                    Estimate  Std.Err  z-value  P(>|z|)   Std.lv  Std.all
##   perc_merit1 =~                                                        
##     prc_f_1 (lmb1)    1.313    0.010  126.406    0.000    0.958    0.695
##     prc_t_1           1.013    0.060   16.823    0.000    0.739    0.636
##   perc_nmerit1 =~                                                       
##     prc_w_1 (lmb1)    1.313    0.010  126.406    0.000    1.134    0.808
##     prc_n_1           1.221    0.050   24.307    0.000    1.054    0.834
##   pref_merit1 =~                                                        
##     prf_f_1 (lmb1)    1.313    0.010  126.406    0.000    1.080    0.774
##     prf_t_1           0.880    0.050   17.469    0.000    0.724    0.596
##   pref_nmerit1 =~                                                       
##     prf_w_1 (lmb1)    1.313    0.010  126.406    0.000    1.019    0.772
##     prf_n_1           1.137    0.051   22.487    0.000    0.882    0.766
##   perc_merit2 =~                                                        
##     prc_f_2 (lmb1)    1.313    0.010  126.406    0.000    0.780    0.627
##     prc_t_2           1.091    0.068   16.033    0.000    0.648    0.593
##   perc_nmerit2 =~                                                       
##     prc_w_2 (lmb1)    1.313    0.010  126.406    0.000    0.979    0.736
##     prc_n_2           1.130    0.070   16.099    0.000    0.843    0.675
##   pref_merit2 =~                                                        
##     prf_f_2 (lmb1)    1.313    0.010  126.406    0.000    0.737    0.632
##     prf_t_2           1.301    0.075   17.407    0.000    0.730    0.648
##   pref_nmerit2 =~                                                       
##     prf_w_2 (lmb1)    1.313    0.010  126.406    0.000    0.766    0.657
##     prf_n_2           1.147    0.068   16.985    0.000    0.669    0.625
##   perc_merit3 =~                                                        
##     prc_f_3 (lmb1)    1.313    0.010  126.406    0.000    0.833    0.692
##     prc_t_3           1.005    0.058   17.346    0.000    0.638    0.621
##   perc_nmerit3 =~                                                       
##     prc_w_3 (lmb1)    1.313    0.010  126.406    0.000    1.005    0.774
##     prc_n_3           1.126    0.060   18.832    0.000    0.862    0.692
##   pref_merit3 =~                                                        
##     prf_f_3 (lmb1)    1.313    0.010  126.406    0.000    0.713    0.649
##     prf_t_3           1.230    0.071   17.380    0.000    0.668    0.617
##   pref_nmerit3 =~                                                       
##     prf_w_3 (lmb1)    1.313    0.010  126.406    0.000    0.886    0.761
##     prf_n_3           1.154    0.055   21.051    0.000    0.779    0.718
## 
## Covariances:
##                    Estimate  Std.Err  z-value  P(>|z|)   Std.lv  Std.all
##   perc_merit1 ~~                                                        
##     perc_nmerit1      0.010    0.023    0.414    0.679    0.015    0.015
##     pref_merit1       0.302    0.027   11.370    0.000    0.503    0.503
##     pref_nmerit1      0.150    0.023    6.628    0.000    0.265    0.265
##     perc_merit2       0.266    0.022   11.889    0.000    0.615    0.615
##     perc_nmerit2     -0.085    0.023   -3.768    0.000   -0.156   -0.156
##     pref_merit2       0.101    0.018    5.488    0.000    0.246    0.246
##     pref_nmerit2      0.092    0.019    4.860    0.000    0.216    0.216
##     perc_merit3       0.242    0.022   11.119    0.000    0.523    0.523
##     perc_nmerit3     -0.084    0.022   -3.727    0.000   -0.150   -0.150
##     pref_merit3       0.052    0.018    2.974    0.003    0.132    0.132
##     pref_nmerit3      0.123    0.020    6.130    0.000    0.249    0.249
##   perc_nmerit1 ~~                                                       
##     pref_merit1       0.388    0.028   13.904    0.000    0.546    0.546
##     pref_nmerit1     -0.077    0.023   -3.350    0.001   -0.114   -0.114
##     perc_merit2      -0.066    0.021   -3.229    0.001   -0.129   -0.129
##     perc_nmerit2      0.235    0.024    9.620    0.000    0.364    0.364
##     pref_merit2       0.134    0.019    6.974    0.000    0.276    0.276
##     pref_nmerit2     -0.068    0.019   -3.512    0.000   -0.136   -0.136
##     perc_merit3      -0.042    0.021   -2.018    0.044   -0.076   -0.076
##     perc_nmerit3      0.295    0.025   11.851    0.000    0.446    0.446
##     pref_merit3       0.135    0.019    7.231    0.000    0.288    0.288
##     pref_nmerit3     -0.082    0.020   -4.006    0.000   -0.140   -0.140
##   pref_merit1 ~~                                                        
##     pref_nmerit1     -0.118    0.024   -4.900    0.000   -0.184   -0.184
##     perc_merit2       0.049    0.021    2.285    0.022    0.100    0.100
##     perc_nmerit2      0.072    0.024    2.995    0.003    0.118    0.118
##     pref_merit2       0.215    0.021   10.324    0.000    0.466    0.466
##     pref_nmerit2     -0.047    0.020   -2.343    0.019   -0.099   -0.099
##     perc_merit3       0.092    0.022    4.243    0.000    0.176    0.176
##     perc_nmerit3      0.090    0.024    3.753    0.000    0.143    0.143
##     pref_merit3       0.204    0.020   10.148    0.000    0.456    0.456
##     pref_nmerit3     -0.034    0.021   -1.595    0.111   -0.061   -0.061
##   pref_nmerit1 ~~                                                       
##     perc_merit2       0.101    0.019    5.195    0.000    0.219    0.219
##     perc_nmerit2     -0.006    0.022   -0.264    0.791   -0.010   -0.010
##     pref_merit2      -0.051    0.018   -2.895    0.004   -0.117   -0.117
##     pref_nmerit2      0.327    0.022   14.825    0.000    0.722    0.722
##     perc_merit3       0.104    0.020    5.310    0.000    0.211    0.211
##     perc_nmerit3     -0.025    0.022   -1.181    0.238   -0.043   -0.043
##     pref_merit3      -0.059    0.017   -3.421    0.001   -0.139   -0.139
##     pref_nmerit3      0.334    0.022   15.010    0.000    0.638    0.638
##   perc_merit2 ~~                                                        
##     perc_nmerit2     -0.008    0.019   -0.392    0.695   -0.017   -0.017
##     pref_merit2       0.195    0.018   10.914    0.000    0.585    0.585
##     pref_nmerit2      0.097    0.017    5.806    0.000    0.279    0.279
##     perc_merit3       0.261    0.020   12.897    0.000    0.693    0.693
##     perc_nmerit3     -0.065    0.019   -3.356    0.001   -0.144   -0.144
##     pref_merit3       0.059    0.015    3.831    0.000    0.183    0.183
##     pref_nmerit3      0.098    0.017    5.630    0.000    0.243    0.243
##   perc_nmerit2 ~~                                                       
##     pref_merit2       0.195    0.019   10.028    0.000    0.465    0.465
##     pref_nmerit2      0.039    0.018    2.148    0.032    0.091    0.091
##     perc_merit3      -0.066    0.020   -3.335    0.001   -0.139   -0.139
##     perc_nmerit3      0.285    0.024   11.887    0.000    0.500    0.500
##     pref_merit3       0.092    0.017    5.231    0.000    0.226    0.226
##     pref_nmerit3     -0.032    0.019   -1.677    0.094   -0.064   -0.064
##   pref_merit2 ~~                                                        
##     pref_nmerit2      0.025    0.015    1.706    0.088    0.077    0.077
##     perc_merit3       0.097    0.016    5.974    0.000    0.272    0.272
##     perc_nmerit3      0.068    0.018    3.821    0.000    0.158    0.158
##     pref_merit3       0.198    0.016   12.190    0.000    0.649    0.649
##     pref_nmerit3     -0.008    0.016   -0.547    0.584   -0.022   -0.022
##   pref_nmerit2 ~~                                                       
##     perc_merit3       0.077    0.017    4.624    0.000    0.207    0.207
##     perc_nmerit3      0.032    0.018    1.729    0.084    0.071    0.071
##     pref_merit3      -0.002    0.014   -0.122    0.903   -0.006   -0.006
##     pref_nmerit3      0.287    0.020   14.702    0.000    0.729    0.729
##   perc_merit3 ~~                                                        
##     perc_nmerit3     -0.043    0.020   -2.200    0.028   -0.088   -0.088
##     pref_merit3       0.201    0.017   11.529    0.000    0.582    0.582
##     pref_nmerit3      0.129    0.018    7.291    0.000    0.302    0.302
##   perc_nmerit3 ~~                                                       
##     pref_merit3       0.187    0.018   10.117    0.000    0.450    0.450
##     pref_nmerit3      0.086    0.019    4.415    0.000    0.166    0.166
##   pref_merit3 ~~                                                        
##     pref_nmerit3     -0.016    0.015   -1.059    0.290   -0.044   -0.044
## 
## Intercepts:
##                    Estimate  Std.Err  z-value  P(>|z|)   Std.lv  Std.all
##    .prc_ffr_1 (i1)    3.297    0.020  164.061    0.000    3.297    2.393
##    .prc_tln_1         3.066    0.029  106.525    0.000    3.066    2.638
##    .prc_wpr_1 (i1)    3.297    0.020  164.061    0.000    3.297    2.349
##    .prc_ntw_1         3.487    0.028  126.736    0.000    3.487    2.757
##    .prf_ffr_1 (i1)    3.297    0.020  164.061    0.000    3.297    2.362
##    .prf_tln_1         3.031    0.030  102.363    0.000    3.031    2.495
##    .prf_wpr_1 (i1)    3.297    0.020  164.061    0.000    3.297    2.500
##    .prf_ntw_1         2.730    0.026  103.746    0.000    2.730    2.369
##    .prc_ffr_2 (i1)    3.297    0.020  164.061    0.000    3.297    2.651
##    .prc_tln_2         3.105    0.036   86.433    0.000    3.105    2.840
##    .prc_wpr_2 (i1)    3.297    0.020  164.061    0.000    3.297    2.480
##    .prc_ntw_2         3.345    0.037   90.001    0.000    3.345    2.679
##    .prf_ffr_2 (i1)    3.297    0.020  164.061    0.000    3.297    2.829
##    .prf_tln_2         2.664    0.048   55.509    0.000    2.664    2.365
##    .prf_wpr_2 (i1)    3.297    0.020  164.061    0.000    3.297    2.828
##    .prf_ntw_2         2.919    0.043   67.281    0.000    2.919    2.728
##    .prc_ffr_3 (i1)    3.297    0.020  164.061    0.000    3.297    2.740
##    .prc_tln_3         3.043    0.033   91.359    0.000    3.043    2.962
##    .prc_wpr_3 (i1)    3.297    0.020  164.061    0.000    3.297    2.540
##    .prc_ntw_3         3.360    0.035   97.079    0.000    3.360    2.698
##    .prf_ffr_3 (i1)    3.297    0.020  164.061    0.000    3.297    3.003
##    .prf_tln_3         2.742    0.050   55.151    0.000    2.742    2.533
##    .prf_wpr_3 (i1)    3.297    0.020  164.061    0.000    3.297    2.830
##    .prf_ntw_3         3.002    0.040   75.414    0.000    3.002    2.768
##     perc_mrt1         0.000                               0.000    0.000
##     perc_mrt2         0.077    0.027    2.908    0.004    0.130    0.130
##     perc_mrt3         0.186    0.026    7.227    0.000    0.293    0.293
##     prc_nmrt1         0.000                               0.000    0.000
##     prc_nmrt2         0.124    0.029    4.295    0.000    0.167    0.167
##     prc_nmrt3        -0.001    0.028   -0.043    0.966   -0.002   -0.002
##     pref_mrt1         0.000                               0.000    0.000
##     pref_mrt2         0.378    0.026   14.769    0.000    0.673    0.673
##     pref_mrt3         0.476    0.025   19.175    0.000    0.876    0.876
##     prf_nmrt1         0.000                               0.000    0.000
##     prf_nmrt2        -0.378    0.024  -15.616    0.000   -0.649   -0.649
##     prf_nmrt3        -0.427    0.024  -17.828    0.000   -0.633   -0.633
## 
## Variances:
##                    Estimate  Std.Err  z-value  P(>|z|)   Std.lv  Std.all
##    .perc_effort_w1    0.981    0.062   15.727    0.000    0.981    0.517
##    .perc_talent_w1    0.805    0.042   18.951    0.000    0.805    0.596
##    .perc_wpart_w1     0.685    0.053   12.871    0.000    0.685    0.348
##    .perc_netw_w1      0.488    0.044   11.017    0.000    0.488    0.305
##    .pref_effort_w1    0.782    0.066   11.911    0.000    0.782    0.401
##    .pref_talent_w1    0.951    0.044   21.535    0.000    0.951    0.644
##    .pref_wpart_w1     0.702    0.048   14.782    0.000    0.702    0.404
##    .pref_netw_w1      0.550    0.036   15.216    0.000    0.550    0.414
##    .perc_effort_w2    0.938    0.051   18.465    0.000    0.938    0.607
##    .perc_talent_w2    0.775    0.038   20.137    0.000    0.775    0.649
##    .perc_wpart_w2     0.810    0.063   12.792    0.000    0.810    0.458
##    .perc_netw_w2      0.850    0.052   16.344    0.000    0.850    0.545
##    .pref_effort_w2    0.816    0.042   19.295    0.000    0.816    0.600
##    .pref_talent_w2    0.735    0.040   18.482    0.000    0.735    0.580
##    .pref_wpart_w2     0.773    0.043   17.835    0.000    0.773    0.569
##    .pref_netw_w2      0.697    0.036   19.446    0.000    0.697    0.609
##    .perc_effort_w3    0.754    0.046   16.332    0.000    0.754    0.521
##    .perc_talent_w3    0.648    0.032   20.090    0.000    0.648    0.615
##    .perc_wpart_w3     0.675    0.054   12.439    0.000    0.675    0.401
##    .perc_netw_w3      0.808    0.046   17.407    0.000    0.808    0.521
##    .pref_effort_w3    0.698    0.038   18.416    0.000    0.698    0.579
##    .pref_talent_w3    0.726    0.036   20.016    0.000    0.726    0.619
##    .pref_wpart_w3     0.572    0.039   14.826    0.000    0.572    0.421
##    .pref_netw_w3      0.569    0.033   17.421    0.000    0.569    0.484
##     perc_merit1       0.532    0.044   12.096    0.000    1.000    1.000
##     perc_nmerit1      0.746    0.045   16.522    0.000    1.000    1.000
##     pref_merit1       0.677    0.049   13.783    0.000    1.000    1.000
##     pref_nmerit1      0.602    0.040   15.182    0.000    1.000    1.000
##     perc_merit2       0.353    0.033   10.627    0.000    1.000    1.000
##     perc_nmerit2      0.556    0.045   12.392    0.000    1.000    1.000
##     pref_merit2       0.315    0.028   11.229    0.000    1.000    1.000
##     pref_nmerit2      0.340    0.030   11.441    0.000    1.000    1.000
##     perc_merit3       0.403    0.033   12.101    0.000    1.000    1.000
##     perc_nmerit3      0.586    0.042   14.060    0.000    1.000    1.000
##     pref_merit3       0.295    0.026   11.403    0.000    1.000    1.000
##     pref_nmerit3      0.456    0.031   14.481    0.000    1.000    1.000

Invarianza Longitudinal en lavaan https://quantdev.ssri.psu.edu/tutorials/intro-basics-longitudinal-measurement-invariance
https://users.ugent.be/~yrosseel/lavaan/multiplegroup6Dec2012.pdf
Invarianza con indicadores categóricos https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5121102/pdf/nihms757737.pdf
Parametrización tetha: https://groups.google.com/forum/#!topic/lavaan/nfdatPgLLhc
Colapsing categories: https://www.tandfonline.com/doi/abs/10.1080/10705511.2018.1547640?journalCode=hsem20

Analysis

20 marzo, 2020

1 Análisis PPM-S

2 Wave 01

2.1 Version 01: Fixed order by perception / preferences (N=697)

2.2 Version 02: Fixed order, by topic (N=706)

2.3 Version 03: Random order (N=697)

2.4 Version 04: muestra completa (N=2236)

2.4.1 Resumen ajuste ola 01

2.5 Análisis por Grupos múltiples

3 Modelo Factorial de Segundo Orden:

3.1 a. Factor de segundo orden por Percepción y Preferencia

3.2 b. Factor de segundo orden por Meritocracia y No Meritocracia

4 Wave 02

4.1 ATE indicadores meritocracia

4.2 CFA Muestra completa (Control, Desigualdad y Pobreza)

4.2.1 Factor de segundo orden

4.3 CFA Muestra Control

4.4 CFA Muestra Pobreza

4.5 CFA Muestra Desigualdad

5 Resumen ajuste ola 2

6 Wave 03

7 Resumen ajuste ola 3

8 Análisis longitudinal

8.1 Correlación indicadores w01 y w02

8.1.1 Grupo Versión 01

8.1.2 Grupo Versión 02

8.1.3 Grupo Versión 03

8.1.4 Muestra completa

8.2 CFA wave01, wave02 y wave03

9 Invarianza